I actually have tried the DIY loop with the older insecure Medtronic pumps that could be controlled over radio. It was openAPS. I'm not sure if there are others

I actually have tried the DIY loop with the older insecure Medtronic pumps that could be controlled over radio. It was openAPS. I'm not sure if there are others or how they work. It ended up being fairly inconvenient. My setup was a pi zero with a radio bonnet running on my desk; it would connect to the internet to pull CGM data, and deliver boluses accordingly. I think the main issue was that I wasn't babysitting it enough to know if it was working or not. It was hard to see/understand bolus decisions it was making unless I checked iob, and it absolutely destroyed the battery life of my pump; I used a AAA once every couple days when I was near my desk, nevermind the battery on the rig (I had it plugged into the wall). Overall, the system that you actually use the most, that is the most reliable, is the most effective system, even if it's a little bit suboptimal. In my case, CIQ is probably a bit less aggressive than the openAPS system; but importantly, it is truly "set and forget." I never have to worry if it's enabled or when, or how, it's just always working, 24/7. With a condition like ours, consistency is the best predictor of performance. That makes it the clear winner imho, unless you really like to tinker with things and you want to be very active with your loop. The other thing is that the software (openAPS) was Javascript and Bash on a Linux system. Those architecture choices just make me nervous. Too many bad things that could happen. I'm not saying it's impossible for open source software to be reliable, certainly not, but the complexity of their system was just too large of a surface area to fully test. There are extremely significant cybersecurity risks using those older pumps also. It's basically root access to the bolus button.